US Patent 9299060 Automatically suggesting groups based on past user interaction

An example implementation includes processing message-related information of a first user for a cluster of related users and determining interaction levels for the cluster of related users. The interaction levels describe a level of interaction between the first user and the related users of the cluster. The users of the cluster are ranked based on the interaction levels and a group suggestion is generated suggesting a group be created on a social networking service based on the ranking of the cluster of related users. The group suggestion is sent to the first user for presentation and a response is received from the first user including user input provided in response to the group suggestion being sent. A group of users is created on the social networking service based on the response.